%0 Conference Paper %1 1220218 %A McClosky, David %A Charniak, Eugene %A Johnson, Mark %B ACL-44: Proceedings of the 21st International Conference on Computational Linguistics and the 44th annual meeting of the Association for Computational Linguistics %C Morristown, NJ, USA %D 2006 %I Association for Computational Linguistics %K parser semisupervised %P 337--344 %R http://dx.doi.org/10.3115/1220175.1220218 %T Reranking and self-training for parser adaptation %U http://portal.acm.org/citation.cfm?id=1220218&dl=GUIDE, %X Statistical parsers trained and tested on the Penn Wall Street Journal (WSJ) treebank have shown vast improvements over the last 10 years. Much of this improvement, however, is based upon an ever-increasing number of features to be trained on (typically) the WSJ treebank data. This has led to concern that such parsers may be too finely tuned to this corpus at the expense of portability to other genres. Such worries have merit. The standard "Charniak parser" checks in at a labeled precision-recall f-measure of 89.7% on the Penn WSJ test set, but only 82.9% on the test set from the Brown treebank corpus.This paper should allay these fears. In particular, we show that the reranking parser described in Charniak and Johnson (2005) improves performance of the parser on Brown to 85.2%. Furthermore, use of the self-training techniques described in (McClosky et al., 2006) raise this to 87.8% (an error reduction of 28%) again without any use of labeled Brown data. This is remarkable since training the parser and reranker on labeled Brown data achieves only 88.4%.

@inproceedings{1220218, abstract = {Statistical parsers trained and tested on the Penn Wall Street Journal (WSJ) treebank have shown vast improvements over the last 10 years. Much of this improvement, however, is based upon an ever-increasing number of features to be trained on (typically) the WSJ treebank data. This has led to concern that such parsers may be too finely tuned to this corpus at the expense of portability to other genres. Such worries have merit. The standard "Charniak parser" checks in at a labeled precision-recall f-measure of 89.7% on the Penn WSJ test set, but only 82.9% on the test set from the Brown treebank corpus.This paper should allay these fears. In particular, we show that the reranking parser described in Charniak and Johnson (2005) improves performance of the parser on Brown to 85.2%. Furthermore, use of the self-training techniques described in (McClosky et al., 2006) raise this to 87.8% (an error reduction of 28%) again without any use of labeled Brown data. This is remarkable since training the parser and reranker on labeled Brown data achieves only 88.4%.}, added-at = {2009-04-17T10:44:05.000+0200}, address = {Morristown, NJ, USA}, author = {McClosky, David and Charniak, Eugene and Johnson, Mark}, biburl = {https://www.bibsonomy.org/bibtex/2dc066983abb39a4dec4466ce2ea2853c/jamesh}, booktitle = {ACL-44: Proceedings of the 21st International Conference on Computational Linguistics and the 44th annual meeting of the Association for Computational Linguistics}, description = {Reranking and self-training for parser adaptation}, doi = {http://dx.doi.org/10.3115/1220175.1220218}, interhash = {913eb75b3c60c27e212fb56b06b78b02}, intrahash = {dc066983abb39a4dec4466ce2ea2853c}, keywords = {parser semisupervised}, location = {Sydney, Australia}, pages = {337--344}, publisher = {Association for Computational Linguistics}, timestamp = {2009-04-17T10:44:05.000+0200}, title = {Reranking and self-training for parser adaptation}, url = {http://portal.acm.org/citation.cfm?id=1220218&dl=GUIDE,}, year = 2006 }